Coda File System

Re: What types of stability problems do people encounter with CODA these days?

From: Stephen J. Turnbull <turnbull_at_sk.tsukuba.ac.jp>
Date: Wed, 6 Oct 1999 09:50:24 +0900 (JST)
>>>>> "Bill" == Bill Gribble <grib_at_cs.utexas.edu> writes:

    Bill> The last buglet before upgrading to 5.3.1 was that a user's
    Bill> home directory (each user's $HOME is a separate Coda volume)
    Bill> somehow got in a bad state.  The clients would hang on any
    Bill> attempt to even "ls" that directory or the directory
    Bill> containing its mountpoint.  AFAICT the server thought
    Bill> everything was just fine.  I shut everything down, upgraded,
    Bill> and everybody's $HOME was back.

In 5.2.x (and maybe 5.0.x? I didn't keep accurate records, I know I
got wedged a few times but not how) I was able to wedge the client in
that way by mv'ing directories.  Shutting down the server and
reinitializing venus fixed the problem.  AFAIK there was no file data
loss, but there possibly could have been, since the directory
information on the server did not get updated to reflect the mv.  So I 
would guess any changes not yet flushed to the server would have been
lost (client restart wasn't enough; I don't know coda well enough to
know if there was a non-scorched-earth way to recover functionality
without reinitializing the cache).

5.3.1 seemed to fix the problem in light testing, but since the "great
SCSI host implosion" of Sept 12 my disks have been relocated to a
different host and my system has not yet been reconfigured to look for
the SCM there, so I don't know much more.

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
__________________________________________________________________________
__________________________________________________________________________
What are those two straight lines for?  "Free software rules."
Received on 1999-10-05 20:51:50